Documentation

Documentation

Stop unsafe agent actions and prove outcomes. Pre-execution authorization and post-execution verification for AI agents.

Two paths: Authority (policy, mandates, sidecar) and Runtime (snapshots, assertions, traces).

I want to…

Control plane: Predicate Vault — centralized policy, kill-switches, and WORM audit for compliance teams.

Pre-execution: policy, mandates, audit

Identity controls, policy enforcement, mandates, revocation, and operational governance.

Agent Security

Zero-trust security wrapper for AI agents with pre- and post-execution verification

Enterprise Governance

Production governance patterns: policy controls, revocation, and audit

Post-execution: snapshots, assertions, traces

Runtime assertions, snapshots, debugging, integrations, and verification APIs.

Choose Your SDK Path

Authority SDK Path

Start with identity, governance controls, and sidecar operations.

Overlay Runtime SDK Path

Start with deterministic runtime checks, traces, and framework integrations.

Need Help?

Can't find what you're looking for? Our support team is here to help.